View
8
Download
1
Category
Preview:
Citation preview
University of Tennessee, KnoxvilleTrace: Tennessee Research and CreativeExchange
Doctoral Dissertations Graduate School
12-2011
Packaging Design of IGBT Power Module UsingNovel Switching CellsShengnan Lisli14@utk.edu
This Dissertation is brought to you for free and open access by the Graduate School at Trace: Tennessee Research and Creative Exchange. It has beenaccepted for inclusion in Doctoral Dissertations by an authorized administrator of Trace: Tennessee Research and Creative Exchange. For moreinformation, please contact trace@utk.edu.
Recommended CitationLi, Shengnan, "Packaging Design of IGBT Power Module Using Novel Switching Cells. " PhD diss., University of Tennessee, 2011.https://trace.tennessee.edu/utk_graddiss/1205
https://trace.tennessee.eduhttps://trace.tennessee.eduhttps://trace.tennessee.edu/utk_graddisshttps://trace.tennessee.edu/utk-gradmailto:trace@utk.edu
To the Graduate Council:
I am submitting herewith a dissertation written by Shengnan Li entitled "Packaging Design of IGBTPower Module Using Novel Switching Cells." I have examined the final electronic copy of thisdissertation for form and content and recommend that it be accepted in partial fulfillment of therequirements for the degree of Doctor of Philosophy, with a major in Electrical Engineering.
Leon M. Tolbert, Major Professor
We have read this dissertation and recommend its acceptance:
Fred Wang, Benjamin J. Blalock, Rao V. Arimilli
Accepted for the Council:Carolyn R. Hodges
Vice Provost and Dean of the Graduate School
(Original signatures are on file with official student records.)
i
Packaging Design of IGBT Power Module
Using Novel Switching Cells
A Dissertation Presented for the
Doctor of Philosophy Degree
The University of Tennessee, Knoxville
Shengnan Li
December 2011
ii
Acknowledgements
First and foremost, I would like to express my great appreciation to my major advisor Dr.
Leon Tolbert. He is a very devoted professor, very supportive advisor, and very nice person. The
opportunity he provided is one of the best. Many thanks to his guidance, supervision in my
research, and help in my life.
My sincere gratitude is also to Dr. Fred Wang. His knowledge and insight helped me a lot.
The discussions with Dr. Wang were always inspiring. His guidance makes great contribution in
my research.
Special thanks to Dr. Zhenxian Liang, Dr. Puqi Ning, Dr. Burak Ozpinezi and Madhu
Sudhan Chinthavali from the National Transportation Research Center. Dr. Liang is very
experienced and helpful. Without his help, I couldn‟t tackle the many challenges in my work.
Puqi offered generous help on module fabrication. Whenever I have a question about module
packaging, I can always count on him. Burak and Madhu are always supportive. I benefited a lot
from their help.
It is a great pleasure to be a member of the UT Power and Energy Lab, which is now
CURENT. The academic and friendly atmosphere is desirable. I will always remember the
inspiring discussions with the students, from which I have learned a lot. I enjoyed the happy time
we spend together.
My thanks are also to Ming Li and my parents. Their endless love is the origin of my
happiness.
iii
Abstract
Parasitic inductance in power modules generates voltage spikes and current ringing during
switching which cause extra stress in power electronic devices, increase electromagnetic
interference (EMI), and degrade the performance of the power converter system. As newer
power devices have faster switching speeds and higher power ratings, the effect of the parasitic
inductance of the power module is more pronounced. This dissertation proposes a novel
packaging method for power electronics modules based on the concepts of novel switching cells:
P-cell and N-cell. It can reduce the stray inductance in the current commutation path in a phase-
leg module and hence improve the switching behavior.
Taking an insulated gate bipolar transistor (IGBT) as an example, two phase-leg modules,
specifically a conventional module and a P-cell and N-cell based module were designed. Using
Ansoft Q3D Extractor, electromagnetic simulation was carried out to extract the stray inductance
from the two modules. An ABB 1200 V / 75 A IGBT model and a diode model were built for
simulation study. Circuit parasitics were extracted and modeled. Switching behavior with
different package parasitics was studied based on the Saber simulation.
Two prototype phase-leg modules were fabricated. The parasitics were measured using a
precision impedance analyzer. The measurement results agree with the simulation very well. A
double pulse tester was built in laboratory. Several approaches were used to reduce the circuit
and measuring parasitics. From the switching characteristics of the two modules, it was verified
that the larger stray inductance in the layout causes higher voltage overshoot during turn off,
which in turn increases the turn off losses.
Multichip (two in parallel) IGBT modules applying novel switching cells was also designed.
iv
The parasitics were extracted and compared to a conventional design. The overall loop
inductance was reduced in the proposed module. However, the mismatch of the paralleled
branches was larger.
v
TABLE OF CONTENTS
Chapter Page
1 Introduction ............................................................................................................................. 1
1.1 Background ...................................................................................................................... 1
1.2 Motivation and Strategy ................................................................................................... 4
1.3 Dissertation Outline.......................................................................................................... 6
2 Literature Review .................................................................................................................... 7
2.1 Wire Bonding Technology ............................................................................................... 7
2.1.1 Material Selection .................................................................................................. 7
2.1.2 Fabrication Process .............................................................................................. 13
2.2 Study on Parasitics Induced by Packaging ..................................................................... 14
3 Layout Design of IGBT Phase-leg Module ........................................................................... 21
3.1 Introduction of P-cell and N-cell .................................................................................... 21
3.1.1 Definition of P-cell and N-cell ............................................................................ 21
3.1.2 DC-DC Converters Constructed from P-cell and N-cell ..................................... 24
3.1.3 Constructing Voltage Source Inverters from the P-cell and N-cell ..................... 30
3.2 Package Layout Design Using the Novel Switching Cells ............................................ 32
3.2.1 Electromagnetic Simulation Methodology .......................................................... 34
3.2.2 Layout Design Using Q3D Extractor .................................................................. 35
3.2.3 Simulation Results ............................................................................................... 38
3.3 Summary ........................................................................................................................ 40
4 Electrical Evaluation Based on Simulation ........................................................................... 41
4.1 Parasitics Extraction from PCB...................................................................................... 41
4.2 IGBT and Diode Device Modeling ................................................................................ 43
4.2.1 Modeling of Diode 5SLY12F1200 ...................................................................... 44
vi
4.2.2 Modeling of IGBT 5SMY12J1280 ...................................................................... 46
4.3 Switching Characterization ............................................................................................ 49
4.4 Modeling and Analysis of Switching Behaviors ............................................................ 52
4.5 Summary ........................................................................................................................ 54
5 Experimental Verification ..................................................................................................... 55
5.1 Parasitics Measurement .................................................................................................. 55
5.2 Static Characteristics Testing ......................................................................................... 61
5.3 Switching Characterization ............................................................................................ 64
5.3.1 Double Pulse Tester ............................................................................................. 65
5.3.2 On Board Inductance Calibration ........................................................................ 70
5.3.3 IGBT Die Voltage Measurement ........................................................................ 75
5.3.4 Experimental Results ........................................................................................... 77
5.3.5 Comparison of Experiment with Simulation results ........................................... 81
5.4 Summary ........................................................................................................................ 83
6 Utilizing Novel Switching Cells in Multichip IGBT Module ............................................... 85
6.1 Layout Design of Multichip Power Modules Based on Novel Switching Cells ............ 85
6.2 Switching Characteristic of Multichip Power Module ................................................... 89
6.3 Measurement of the Parasitics in the Multichip Power Modules................................... 92
6.4 Summary ........................................................................................................................ 97
7 Conclusion and Future Work ................................................................................................. 98
7.1 Conclusion ...................................................................................................................... 98
7.1.1 Summary of the Work ......................................................................................... 98
7.1.2 Power Module Design Considerations and Influence of Parasitics .................. 100
7.1.3 Contributions of This Dissertation .................................................................... 101
7.2 Future Work ................................................................................................................. 102
vii
References ................................................................................................................................... 104
Vita. ............................................................................................................................................. 111
viii
LIST OF FIGURES
Figure 1.1. SEMiX product family. ................................................................................................ 2
Figure 1.2. 1200 A, 3300 V IGBT module. .................................................................................... 2
Figure 1.3. Fully integrated intelligent power module in single-in-line and dual-in-line housing. 3
Figure 1.4. SKiM63 (300 A, 1200 V) and HybridPack (400 A, 600 V)......................................... 3
Figure 2.1. Structure of an IGBT power module. ........................................................................... 8
Figure 2.2 Package inductance ..................................................................................................... 19
Figure 3.1. Basic canonical cell. ................................................................................................... 22
Figure 3.2. Two basic switching cells: P-cell and N-cell. ............................................................ 24
Figure 3.3. Four classic DC-DC converters and their P-cell and N-cell representation ............... 25
Figure 3.4. Simulation results for P-cell and N-cell buck converter. ........................................... 26
Figure 3.5. Experimental output voltage ripple (100mV/div) of buck converter. ........................ 27
Figure 3.6. Experimental output voltage of Ćuk converters. ........................................................ 30
Figure 3.8. Full bridge inverter with package parasitics. .............................................................. 33
Figure 3.9. Phase-leg module layout............................................................................................. 37
Figure 3.10. Phase-leg modules with measuring points. .............................................................. 39
Figure 4.1. PCB used for double pulse tester................................................................................ 42
Figure 4.2. DC link in double pulse tester PCB. ........................................................................... 43
Figure 4.4. Saber diode model user interface. .............................................................................. 45
ix
Figure 4.5. Scanned I-V curve from datasheet. ............................................................................ 46
Figure 4.6. I-V characteristic after curve fitting from the loaded I-V curve. ............................... 46
Figure 4.7. Saber IGBT model user interface. .............................................................................. 48
Figure 4.8. Verification of gate charge. ........................................................................................ 48
Figure 4.9. Verification of switching waveform. .......................................................................... 49
Figure 4.10. Schematic of double pulse tester. ............................................................................. 50
Figure 4.11 Typical waveforms of double pulse tester. ................................................................ 51
Figure 4.12. Switching behaviors under module parasitics. ......................................................... 52
Figure 4.13. Switching transient circuit model. ............................................................................ 54
Figure 5.1 Conventional module DBC pattern with bonding wires designed for parasitic
measurement ................................................................................................................................. 57
Figure 5.2. Proposed module DBC pattern with bonding wires designed for parasitic
measurement. ................................................................................................................................ 58
Figure 5.3. Impedance analyzer and probe fixture for parasitic measurement. ............................ 59
Figure 5.4. Measured results. ........................................................................................................ 60
Figure 5.5. Fabricated modules (Conventional one in the left and the proposed one in the right)61
Figure 5.6. Tektronix 371B curve tracer. ...................................................................................... 63
Figure 5.7 Measured IGBT output characteristic ......................................................................... 63
Figure 5.8. IGBT output characteristic from datasheet................................................................. 64
Figure 5.9. FWD I-V curve. .......................................................................................................... 64
x
Figure 5.10. Printed circuit board of DPT. ................................................................................... 66
Figure 5.11. Demonstration of the cable stray inductance effect. ................................................ 67
Figure 5.12. Gate drive circuit. ..................................................................................................... 69
Figure 5.13. Reduced tip-to-ground loop using a probe-tip adaptor............................................. 70
Figure 5.14. Measurement result of the bus impedance. .............................................................. 71
Figure 5.15. Measurement results of decoupling capacitance on board. ...................................... 72
Figure 5.16. Stray inductance measurement results of the PCB. .................................................. 73
Figure 5.17. Layout of DC bus in PCB. ........................................................................................ 73
Figure 5.18. Voltage on module terminals during turn off. .......................................................... 74
Figure 5.19. Voltage on module terminals during turn off after correction. ................................ 74
Figure 5.20. Illustration of measuring points. ............................................................................... 76
Figure 5.21. Module with wire connection from IGBT die. ......................................................... 76
Figure 5.22. Turn off voltage of IGBT at different measuring points. ......................................... 77
Figure 5.23. Enlarged waveform of turn off voltage of IGBT at different measuring points. ..... 77
Figure 5.24. Turn off voltage and current. .................................................................................... 78
Figure 5.25. Close-up view for voltage overshoot during turn off. .............................................. 78
Figure 5.26. Slope of the turn off current. .................................................................................... 79
Figure 5.27. Voltage across diode................................................................................................. 80
Figure 5.28. Turn off loss calculation. .......................................................................................... 81
xi
Figure 5.29. Comparison of experiment and simulation of gate voltage (red is experiment; blue is
simulation). ................................................................................................................................... 82
Figure 5.30. Comparison of experiment and simulation of IGBT voltage. .................................. 83
Figure 5.31. Comparison of experiment and simulation of IGBT current.................................... 83
Figure 5.32. Analog circuit representation of IGBT model. ......................................................... 83
Figure 6.1. A commercial inverter IPM. ....................................................................................... 86
Figure 6.2. Conventional layout of an IGBT module with paralleled dice................................... 86
Figure 6.3. Parasitics extraction of the conventional multichip module....................................... 87
Figure 6.4. Layout design of a paralleled IGBT module using novel switching cells. ................. 88
Figure 6.5. Parasitics extraction of the conventional multichip module....................................... 88
Figure 6.6. Turn off voltage of IGBTs. ......................................................................................... 90
Figure 6.7. Turn on current of IGBTs. .......................................................................................... 91
Figure 6.8. Turn on current of diodes. .......................................................................................... 91
Figure 6.9. Turn off current of diodes........................................................................................... 92
Figure 6.10. Commutation loops in the conventional layout. ....................................................... 94
Figure 6.11. Commutation loops in the proposed layout. ............................................................. 95
Figure 6.12. Fabricated commutation loops in the conventional layout. ...................................... 95
Figure 6.13. Fabricated commutation loops in the proposed layout. ............................................ 96
Figure 6.14. Stray inductance measurement results...................................................................... 96
xii
LIST OF TABLES
Table 2.1. Properties of Insulating Substrate .................................................................................. 9
Table 2.2. Maximum Current for Aluminum Wire Size ............................................................... 12
Table 2.3. Summary of Parasitic Inductance in Power Modules .................................................. 15
Table 2.4. Comparison of Parasitics in Conventional and U-series Power Module ..................... 19
Table 2.5. Summary of Methods to Reduce Inductance ............................................................... 19
Table 3.1. Power Module Components Dimensions .................................................................... 38
Table 3.2. Power Module Materials .............................................................................................. 38
Table 3.3. Simulation Results of the Phase-leg Module Stray Inductance ................................... 40
Table 4.1. Parameters in Double Pulse Tester .............................................................................. 50
Table 5.1. Stray Inductance Comparison of the Measurement and Simulation ............................ 61
Table 5.2. Circuit parameters ........................................................................................................ 65
Table 5.3 Comparison of measured and calculated inductance. ................................................... 79
Table 6.1. Test Condition of DPT ................................................................................................. 89
Table 6.2. Comparison of the simulation and measurement results. ............................................ 97
1
1 Introduction
Power semiconductor modules play a key role in a power electronic system, such as
switching mode power supply, motor drive, UPS, and so on. Generally, a power
semiconductor module may be defined as a device which contains more than one
semiconductor chip and provides an electric path and a heat flux path [1]. The first power
semiconductor module was established in the mid seventies of the past century. For the first
time, two chips were combined by being soldered together with electrical contacts on
metalized ceramic substrates and by being put in a common plastic housing. Power electronic
systems became much more compact, cost efficient and reliable, which necessitated
advanced device packaging and integration technology. As decades went by, power device
and module packaging technology has evolved through multiple generations, each with
incremental improvements in performance and reliability. However, the requirements for
lower cost, small size, light weight and more reliable power modules have not ceased.
1.1 Background
At present, the insulated gate bipolar transistor (IGBT) has become the device of choice
for a wide range of industrial power conversion applications because of the superior
switching characteristics, low losses, and simple gate drive. IGBTs make up 43% of the
power module market [1] with 56% of the IGBT applications being motor drives. Figure 1.1
shows a series of IGBT modules from SEMiX in this application area, it offers rectifiers and
IGBTs for 15 kW to 150 kW in a package outline with standard 17 mm terminal height.
Another application is traction, which is the highest power condition an IGBT can handle.
2
IGBT modules for traction are shown in Figure 1.2 (Photo of ABB Semiconductors).
Intelligent power module (IPM) as shown in Figure 1.3 (Photos of Fairchild Semiconductors,
International Rectifier, Mitsubishi Electric) is another popular application area. Moreover,
two emerging market segments will be highlighted as well. One is power modules for
renewable energy applications, in particular wind power, and the other is automotive
applications. Power modules in wind power are basically similar as in traction. However in
automotive, thermal behavior is critical because of the vehicle environment, hence the IGBT
module packaging is usually of high temperature. As an example, Figure 1.4 shows a
commercial automotive module designed by Semikron.
Figure 1.1. SEMiX product family.
Figure 1.2. 1200 A, 3300 V IGBT module.
3
Figure 1.3. Fully integrated intelligent power module in single-in-line and dual-in-line housing.
Figure 1.4. SKiM63 (300 A, 1200 V) and HybridPack (400 A, 600 V).
High reliability and long term stability are essential in high power applications.
According to [8], a 30-year lifetime, 338,000 long-term cycles, and 12 million short-term
temperature changes are required for traction applications. As stated above, IGBTs are being
extensively used for relatively high current and high frequency applications. A scenario that
is commonly used to explain the IGBT failure is the coefficient of thermal expansion (CTE)
mismatch between the silicon and substrate during thermal cycling. Actually, in application,
some failures are caused by parasitic effects. One example is the voltage spike during
switching, which is a function of total dc loop inductance. It can only be controlled
effectively by the gate resistance of the other commutated IGBT at the expense of high turn-
on loss. After the IGBT fails, it is commonly found, that one or two bonding wires were
opened, or the chip surface at the bonding joints is cracked [19].
Excessive overshoot voltage is harmful to the IGBT safe operation area and even causes
IGBT destruction. In addition to the voltage spikes, another problem is current ringing due to
4
the resonance of the parasitic inductance in the power module and the parasitic capacitance
of the devices. It is a source of electromagnetic interference (EMI) and leads to extra power
loss. Last but not the least is the current sharing problem in multichip modules because of the
discrepancy of the parasitics.
In order to solve the problems mentioned above to the most extent, it is important to
reduce module internal inductance [7]. Much research work has been done to study the
parasitics in the power module, basically the parasitic inductance is distributed in the
terminal leads, bonding wires and substrate.
1.2 Motivation and Strategy
As the demand of power rating and switching speed of the power electronics devices
increases, the current slope, di/dt, is getting larger, the role of the stray inductance inside the
module is more and more important. The objective of this work is to design the packaging
layout based on the concepts of two novel switching cells to reduce the stray inductance to
the most extent.
The dominant technology in power module packaging will still be wire bond technology.
Therefore, this technology is used as the packaging method in this dissertation. Power IGBT
modules will be the focus of discussion owing to its rising popularity. However, the results of
this work are general and can be used in other types of power devices and package
technology as well.
First, the IGBT module will be modeled using electromagnetic field simulation software
Q3D Extractor from Ansoft, which is a 3-D and 2-D parasitic extraction software tool for
designing printed circuit boards (PCB), electronic packaging, and power electronic
5
equipment. Specifically, Method of Moments (integral equations) and Finite Element
Methods were used to compute capacitance, conductance, inductance, and resistance
matrices [9]. With this simulation tool, the design process will be much easier and cost
effective, since the parasitics can be studied in detail before the module prototype is
fabricated.
After the layout design was done and all the parasitics were extracted, circuit simulation
was carried out to characterize the circuit behavior. Synopsys Saber is used as the simulation
tool in this case. Saber was a multi-domain modeling and simulation environment that
enables full-system virtual prototyping for applications in analog/power electronics, electric
power generation/conversion/distribution and mechatronics. Saber has been used for design
validation and optimization for automotive, aerospace, and industrial systems [10]. In this
work, IGBT turn-on and turn-off behaviors were selected as the index for the electrical
evaluation, since the packaging parasitics mainly affect the switching behavior. A double
pulse tester (DPT) which is the standard circuit for switching characterization was used in the
simulation.
After a clear picture of the parasitics in power modules and the circuit behavior under
the module parasitics was obtained from the simulations, real power modules were fabricated
and tested. The module internal parasitics were measured and an experimental DPT was built
and tested to verify the proposed concept and simulation.
The novel switching cells concept was next extended to the multichip power modules.
The same approaches were used to study the effects of the module stray inductance.
6
1.3 Dissertation Outline
According to the strategy discussed above, the outline of the dissertation is listed as
follows.
Chapter 2 is the literature survey. The dominant packaging technology nowadays, wire-
bonding technology, is discussed in the beginning. Then the techniques of layout design and
other considerations in terms of power module electrical behaviors are reviewed.
Chapter 3 presents the layout design of the power module. This layout design originates
from the concepts of P-cell and N-cell. A conventional phase-leg module and the proposed
phase-leg module are built in Q3D extractor.
Chapter 4 is the DPT Saber simulation under the extracted parasitics. The turn-on and
turn-off overshoot voltages and ringing current are compared for two different packages.
Also, analytical modeling and analysis for both turn-on and turn-off is done.
Chapter 5 is the module parasitic measurement and experimental results of DPT.
Chapter 6 discusses stray inductance in modules with paralleled devices. Based on the
parasitics extraction and the circuit simulation, both the advantages and disadvantages of the
proposed design layout is discussed.
Chapter 7 concludes the work that has been done in the dissertation, summarizes the
main contributions and the possible future work.
7
2 Literature Review
Power electronics packaging technology has been developed for several generations,
involving material upgrading, structure improvement, and interconnection technique
innovation. For instance, in high power situations, a press pack is used to reduce bonding
wires and solder attachment. Also, planar technology has gained much interest in certain
applications (such as hybrid electric vehicles) since double sided cooling is enabled. In most
commercial modules, nevertheless, wire bonding is still the main choice of packaging due to
its maturity and reliability.
2.1 Wire Bonding Technology
In general, a power module construction is composed of different layers, as shown in
Figure 2.1 [11]. The base layer is the baseplate, which is a thick layer of metal used for
mechanical fixation and heat transfer. Direct bonded copper (DBC), which consists of two
copper layers and one ceramic layer, is soldered on the baseplate. Power semiconductor dice
and terminals are then soldered on the top layer of the DBC. Moreover, aluminum wires are
used for interconnection of dice and terminals. The module is put in a plastic case. Finally,
encapsulant such as silicone gel is filled in the case for protection and insulation. In fact,
power module fabrication involves many different processes and techniques, and several key
factors of them will be discussed in detail in this chapter.
2.1.1 Material Selection
The materials involved in a power module design cover a wide range from insulators,
conductors, and semiconductors to organics. Since these materials behave differently under
8
various environmental, electrical, and thermal conditions, proper selection of these materials
and the assembly processes are critical.
A. Substrate selection
Substrate as shown in Figure 2.1 is one of the most important parts in a power module.
Typically, there are three layers for the substrate, i.e., two conduction layers and an
insulation layer in between. Specially, the top metal layer is the printed circuit of the power
module. The insulation layer serves as the supporting structure for the circuitry [12]. This
layer is mechanically a base to support all active and passive chip components, and
electrically an insulator to isolate various conductive paths of the circuit. The bottom metal
layer is for thermal expansion balancing.
Figure 2.1. Structure of an IGBT power module.
Two commonly used materials for the substrate are Aluminum Nitride (AlN) and
Alumina (Al2O3). The properties of concern are listed in Table 2.1.
9
Table 2.1. Properties of Insulating Substrate
Material Al2O3 (96%) Al2O3 (99%) AlN
Electrical Resistivity (W-cm) > 1014
> 1014
> 1014
Dielectric Strength (kV/mm) 12 12 15
Dielectric Constant at 1 MHz 9.2 9.9 8.9
Thermal Thermal conductivity (W/m ˚K) 24 33 150-180
CTE (ppm/˚C) 6.0 7.2 4.6
Heat Capacity (J/kg-˚C) 765 765 745
Maximum Use Temperature (˚C ) 1600 1600 >1000
Melting Point (˚C) 2323 2323 2677
Mechanical Tensile Strength (MPa) 127.4 206.9 310
Flexural Strength (MPa) 317 345 360
Density (kg/m3) 3970 3970 3260
Elastic Modulus (GPa) 310.3 345 310
Hardness 2000K 9MH 1200K
Surface Finish (μm) 1.0 1.0 1.0
Others Metalizability All, except thin
film
All All, except thick
film
Machine ability Good Good Good
Relative cost 1 2 4
The top and bottom metallization also have some requirements for power electronics
application. Some of the key requirements are listed below [11]:
– High thermal conductivity (> 200 W/k-m).
–High current density.
– Strong adhesion to the substrate.
– Photoetchable.
According to the requirements above, DBC becomes the most popular choice for the
10
substrate. DBC technology uses a high-temperature process to achieve an intimate bond
between the copper and the ceramic. There is no solder or any other catalyst used in the
interface between the copper and the ceramic surface. Here, the combination of copper and
ceramic is heated to a temperature of about 1070°C which is slightly below copper‟s melting
point, in a nitrogen atmosphere. At this temperature, the copper oxide forms a eutectic melt
that wets and, when cool, produces a strong bond between copper and ceramic. Copper
thickness is typically 8 to 20 mils.
B. Bonding material
Besides substrate, bonding materials also play a key role in a power module. It provides
the vital functions of mechanical, thermal, and electrical linkages between the power
semiconductor chips, the terminals, the insulating substrates, and the metal base plate.
Therefore, bonding must be properly designed to ensure that the power IGBT module is a
mechanically reliable and thermally efficient system.
Solder becomes the bonding choice after comparing with other materials such as epoxy
and silver filled glass. Solders are essentially alloys of two or more metals. When these
metals are alloyed together, the melting point of the alloy can be considerably less than the
melting point of either of the individual starting metals (a phenomenon which makes the
soldering process possible). In the soldering process, the solder is placed between two metal
surfaces to be soldered. During melting, the molten solder dissolves a portion of these two
surfaces and, when the solder cools, a junction or solder joint is formed, joining the two
metal surfaces.
Selection of the solder alloys is based on the following criteria [13]:
11
• Melting temperature range in relation to service temperatures. Due to the CTE
mismatch between the power chip, the insulating substrate, and the metal base plate, the
processing temperature of the solder should be as low as possible and is preferred to be at or
below 350°C. This processing or soldering temperature is typically 20 ºC to 40ºC above the
solder melting temperature. Usually, Tj of the IGBT chip can be as high as 150ºC. Thus, the
solder melting temperature must be at least 10ºC higher to prevent any remelting.
– Processing restrictions. Usually, the power chips are first attached to the insulating
substrate using a high temperature solder. The insulating substrate is then attached to the
metal baseplate with a lower-temperature solder to avoid remelting of the first solder. This is
done so that the two solder attachments can be optimized independently. These two soldering
temperatures should be at least 40ºC apart. According to this, the melting temperature ranges
for the two solders should be as follows: first solder - 200˚C to 310˚C, and second solder -
160˚C to 270˚C.
• Availability. The solder should be available in both preform and paste form.
• Compatibility with the metallization of the power chips, the insulating substrate, and
the metal baseplate.
Also, the criteria include high mechanical strength, low elasticity, high-creep, high-
fatigue resistance and so on.
The most commonly used alloy systems in semiconductor assembly are:
• Gold/Tin hard solder
• Tin/lead soft solder
C. Power interconnection material
12
The interconnection part in a power module is bonding wires that are used between the
top aluminum metallization of the IGBT/Free-wheel diode (FWD) surface and the
electroplated substrate, and terminal leads, which are used to connect to outside circuit.
Aluminum wire is preferred for wire bonding because of its low electrical resistivity,
reliable attachment to the chip metallization surface and substrate, and also its affordable
price. The standards for aluminum wire selection is listed in Table 2.2.
Table 2.2. Maximum Current for Aluminum Wire Size
Material Diameter
(Inch)
Maximum Current (A)
L < 0.040" L > 0.040"
Aluminum (Al/1%Mg) 0.001 0.7 0.5
0.002 2.0 1.4
0.005 7.8 5.4
0.008 15.7 10.9
0.012 28.9 20.0
0.015 40.4 28.0
0.022 71.8 49.6
In terms of terminal connections, the requirements for this material are high conductivity,
high mechanical strength, and elasticity for molding and so on. Copper based alloys such as
beryllium/copper (BeCu, C170, C172) and phosphor/bronze, are commonly used because of
the property of low resistivity, high tensile strength, ease of fabrication and reshape, high
fatigue endurance and wear resistance. The terminals are formed by soldering to the substrate
or by integrating into the case, and connect to the chip or substrate using aluminum wires.
13
2.1.2 Fabrication Process
The procedures of the fabrication process include cleaning the devices, soldering, wire
bonding and encapsulating. This session reviews the major techniques: soldering and wire
bonding.
Devices and terminals soldering
As mentioned briefly earlier, the common method for component attachment in IGBT
module assembly is a two-solder process. First, power chips will be soldered to the substrate
using the first solder. After the chips are put at the correct position on the substrate with the
solder preform, the assembly is moved automatically onto the conveyer belt of the reflow
oven for soldering. Ceramic substrate, power terminal and connecting bridge attachments are
usually done manually by using a graphite fixture or a graphite fixture with alignment sheets.
A graphite fixture is in the shape of a block with detachable parts for the placement of base
plate, ceramic substrate terminals, and bridges. Solder can be either screen-printed or
dispensed onto the baseplate and ceramic substrate. For graphite fixture with alignment
sheets, it is a flat graphite plate about 10 mm thick with guided pins that are in line with the
mounting holes of the baseplate and the sheets. The sheets or frames are metallic, usually
stainless steel of 1 to 4 mm thick, and have windows etched out for the placement of different
components. The guided pins are designed with stops to control the height of these sheets
above the base plate.
The second solder is for attaching the terminals/connecting bridges to the ceramic
substrate and the ceramic substrate to the metal base plate. Generally, the first solder has a
melting temperature of about 25˚C to 40˚C higher than the second solder [11].
14
Ultrasonic wire bonding
After the solder attachment is finished, the interconnection inside the module will be
done using wire bonding. A bonding machine and aluminum wires are involved in this
process. During bonding, the wedge presses the wire against the metal termination pad, and
ultrasonic energy (usually 20 to 60 kHz) is applied to the wedge. The wire is rubbed against
the contact, causing local heating and a metallurgical weld. The thin oxide coating on the
aluminum wire is ruptured, and the oxide tends to help the friction heating process, giving a
very reliable bond.
2.2 Study on Parasitics Induced by Packaging
Parasitic inductance exists from the IGBT chip collector and emitter to their terminal
connections, no matter what kind of packaging technique is used. The parasitic inductance
stores energy whenever the current flows through the interconnections inside the module
when the IGBT is on. When it turns off, the energy is released directly as a voltage spike if
there is no external snubber capacitor in the current loop. This spike is a function of
inductance and di/dt rate. Even with careful layout of the power stage, a snubber capacitor is
usually needed to absorb this energy. If the snubber loop equivalent series resistance (ESR) is
small, a high oscillatory current is produced. If the ESR is large, the current waveform is
improved, at the expense of circuit loss and heating. Although the IGBT is designed to
remove the conventional heavy-duty snubbers such as resistor-capacitor-diode, it cannot
survive without some form of snubber circuit in most applications. How to deal with the
parasitics effect will ultimately affect the EMI, efficiency, and performance of the converter
[19]. In the design and layout of IGBT packages and power stages with both high switching
speed and high power handling requirements, reducing parasitics is extremely important.
15
The package stray inductance can be classified into three categories, as follows [6]:
1. Inductance due to DBC substrate pattern;
2. Inductance due to bonding wires;
3. Inductance of electrode.
The substrate inductance is the smallest among the three. The inductance in the bonding
wires depends on the length. Usually, the length of wire is minimized and hence the induced
inductance is not a big concern. The terminal conductors have a relatively large dimension,
thus the largest inductance exists in this part. Table 2.3 lists the parasitics in a 300 A 1200 V
commercial power module [19], which provides a rough idea of the scale and how the
parasitics are distributed in a power module.
Table 2.3. Summary of Parasitic Inductance in Power Modules
Bonding
wire
Emitter conductor
trace
Collector conductor
trace
Terminal
conductors
Parasitic
inductance
10-15 nH 5-7 nH 4-5 nH 30-40 nH
Although the stray inductance induced by the bonding wires is not the biggest concern in
the module parasitics, there is a special issue with the paralleling of the wires. Table 2.2
shows the current capabilities of the different aluminum wires. 12 mil and 15 mil diameter
wires are most commonly used for IGBT modules. Very often, the length of the wire
connection is longer than 40 mils. Therefore the current capability of a single wire is 20 A for
12 mil wire and 28 A for 15 mil wire. A single IGBT die can have current capability up to
300 A. Considering no overload margin, twenty 15 mil thick wires are needed to parallel for
a 300 A IGBT. The paralleling of the bonding wires can cause the following problems [19]:
1. Proximity Effects between Bonding Wires
16
In the transients where the current rises and falls quickly, the equivalent high frequency
contents are concentrated. At this moment, the mutual coupling effect between the paralleled
wires become so significant that the wires in the edge appear as a low impedance path
compared to the middle wires. Those wires carry more current than the others. Meanwhile,
because of the non-uniform bonding wires on the chip, the steady state current distribution is
affected. This can possibly load the IGBT cells inside the silicon differently. The current
distribution in different wires during transient is illustrated in [19]. The turn on current
overshoot in the wire in the edge can be as large as three times of that in the middle wire.
2. Mechanical Stress on Bonding Wires
It is known that a magnetic force will be generated on a conductor carrying current when
it is exposed to a magnetic field. The force generated on a particular wire can be regarded as:
fi = BiiLi
where, ii and Li, are the current and the length of the i-th wire, and B is the magnetic flux
density at that position The magnetic flux density is the sum of the flux density generated by
all the other bonding wires, if we assume that the wires are perfectly in parallel, and its
length is much longer compared to their distance. The flux density contributed by wire k to i
can be written as:
Bi,k=ikµ0/(2πδi,k)
where δi,k is the distance between the two wises, ik is the current magnitude in wire k. With
this simplified model, the magnetic force generated on the bonding wires can be simulated.
The amplitude of the stress applied to the bonding wires may not be a significant number.
But under temperature change and power cycling, with this repetitive switching frequency
17
lateral force, there is a bonding fatigue mechanism as the wires try to peel the metallization
off the chip surface, which can increase the on-voltage drop of the device. This explains why
the wire-bond modules are more fragile under large repetitive transient current.
High reliability and long term stability are essential in high power applications.
Therefore, reducing package stray inductance is an important issue. There are several
considerations and improvements in the structure of the package to reduce the parasitics of
the module. They are discussed as follows:
1. Terminal arrangement
As stated previously, in a power module the dominant stray inductance is the terminal.
Several methods were proposed to reduce this stray inductance. For example, a laminated
structure has smaller self inductance. Also, when paralleling the positive and negative
terminals, it enables the coupling of the two inductors to the most extent. Theoretically, the
equivalent loop inductance equals the two self-inductances minus the mutual inductance. A
larger mutual inductance will give smaller total equivalent loop inductance. Therefore, when
designing the terminals, the best way is to put two parallel laminated bus bars as close as
possible.
2. Bond wires consideration
First, the interconnection of bond wires should be as short as possible. Second, the
direction of substrate current, which flows under the emitter bonding wires, is designed to be
opposite to the direction of current flow in bonding wires. This wiring construction on the
substrate is also achievable by employing multi-layered DBC technology. Consequently, the
18
bonding wires effect on the module internal inductance can become practically negligible.
3. Utilizing the substrate area
Although the substrate has the smallest inductance, large substrate area can still make
the inductance considerably large. It is especially true for high power modules because the
paralleling of the power devices enlarges the substrate area. To accommodate the bonding
wire connection, the substrate area has to be larger than the footprint of the semiconductor
dice. When doing the substrate layout, maximum utilization of the full substrate area should
be done.
4. The “U-package” technology [6]
Mitsubishi made a major improvement on the bus bar structure in the sense of reduced
stray inductance in 1996. Specifically, the bus bars are molded into the sides of the case,
aluminum wires are used to connect the substrate or die to the terminal. Paralleling the main
electrodes and narrowing the space are easily made. The distance between both electrodes
can be reduced to benefit from the eddy current effect. This also relieves "S" bends that are
needed in the electrodes of conventional modules. Elimination of these "S" bends helped to
further reduce the electrode inductance. Overall, as a result of these inductance reducing
features, the new package has about one third the inductance of conventional modules. Table
2.4 shows the package inductance comparison between conventional and the new concept of
U series.
19
C1
E1C2
E2
Figure 2.2 Package inductance
Table 2.4. Comparison of Parasitics in Conventional and U-series Power Module
Inductance between terminals Conventional U series
C1-E1C2 51 17
E1C2-E2 34 24
C1-E2 58 16
For ease of reference, some general rules to reduce the parasitic inductance are
summarized in Table 2.5 [6].
Table 2.5. Summary of Methods to Reduce Inductance
Classification Inductance reduction methods
DBC substrate pattern 1. Widen the pattern width
2. Shorten the pattern length
Bonding wires 1. Shorten the wire length
2. Increase the number of Al-wires
3. Increase the diameter of the wires
Inductance of electrode 1. Shorten the length
2. Increase the width
3. Parallel the main electrodes and reduce the space
between the main electrodes
4. Use eddy current effect
20
People are making every effort to reduce the stray inductance inside the module.
However, one important issue has been neglected, i.e., the effective stray inductance while
the module is operating. Actually, the stray inductance is everywhere in a module, the focus
is that in the conduction path during switching on and switching off. The following chapter
will introduce concepts of two basic switching cells in power converters, reveal the
mechanism of how the switching cell works as a functional unit, and show how it can affect
power module packaging.
21
3 Layout Design of IGBT Phase-leg Module
As the basic elements, switching devices (mainly MOSFET and IGBT) and diodes along
with inductors and capacitors are used in power electronic circuits to perform dc-dc, dc-ac,
and ac-ac power conversion. In a piece-wise fashion, many circuits have been invented,
proposed, and demonstrated to perform these power conversion uses [20]. The classical dc-
dc converters like buck, boost, buck-boost and Ćuk converters are used in various
applications, and the modeling of these various structures is very important to design the
control circuits for these converters [20]-[28]. However, these circuits have rarely been
examined and investigated in terms of their relationships, topological characteristics and their
basic building blocks. After examining the basic building blocks of these dc-dc converters
and dc-ac inverters, two basic switching cells are proposed [29].
These basic switching cells function as the fundamental elements in power electronic
circuits, which cannot be further broken down or apart and should be used as the basis for
manufacturing/layout of single, dual, and 6-pack modules that semiconductor manufacturers
are producing. Actually, existing well-known circuits can easily be represented and
configured from the basic switching cells. Moreover, some new conversion topologies can be
derived by rearrangement of basic switching cells.
3.1 Introduction of P-cell and N-cell
3.1.1 Definition of P-cell and N-cell
The introduction of the switching cell concept started with the canonical cell [31][32],
where an inductor, a capacitor, and a single-pole double throw switch form a basic canonical
22
switching cell as shown in Figure 3.1. This cell has three terminals A, B and C, and each of
them can be used as an input/output/common terminal. For instance, if terminal A is used as
an input, B as an output and C as the common terminal, the canonical circuit forms one type
of dc-dc converter. Six different combinations can be formed by changing the function of the
three terminals for different combinations [20]. Among these six combinations, only three
distinct effective circuits are found whereas the others are functionally the same. Thus, using
these three combinations, the buck, boost and buck-boost converters can be formed.
L
C
+ vC -
A B
C
S
iy
iz
iL
y z
x
Figure 3.1. Basic canonical cell.
Besides the canonical switching cell, there are many reported methods of modeling
power electronic converters out of some switching modules or blocks. According to [32], the
classical converters can be grouped into two major converter families - buck converter and
boost converter. The buck family converters‟ small signal models can be expressed in terms
of h-parameters, while those for the boost families are expressed by g-parameters. When a
unity feedback is applied in the buck converter, the buck-boost converter is obtained.
23
Using the technique presented in [33], the classical PWM converters can be represented
by only the buck and boost converter connected in cascaded arrangement. This method is
reported as the graft scheme, which presents a unified and systematic method to synthesize
and model transformer-less PWM dc-dc converters. To do that, 4 different basic unit cells
were presented where the cells are made from two transistors. Then using the graft scheme,
the diode-transistor realization of those 4 cells was derived.
Figure 3.2 shows the two basic switching cells defined in this paper. Each cell consists
of one switching device (a MOSFET or IGBT) and one diode connected to three terminals:
(+), (-), and (→) /or (←). Each cell has a common terminal which is shown as (→) /or (←)
on the schematic. For the P-cell, this common terminal is connected to the positive terminal
of a current source or an inductor. For the N-cell, this common terminal is connected to the
negative of a current-source or an inductor. The active switching device in a P-cell is
connected between the (+) and common terminal, whereas in an N-cell, the switching device
is connected between the (-) terminal and the common terminal. Thus, the P-cell is
essentially the mirror circuit of the N-cell and vice versa [29][30].
The aforementioned basic switching cells are the practical implementation of the
canonical switching cell found in [20]. Although the switching cells have only two
components, they can be connected in different combinations to create various power
electronic circuits.
24
Figure 3.2. Two basic switching cells: P-cell and N-cell.
3.1.2 DC-DC Converters Constructed from P-cell and N-cell
Figure 3.3 summarizes the four classical converters and their cell structures. In Figure
3.3, there are three columns and each column has 4 figures. The figures in the left most
column show the four major classical converters. These converters are made from inductors,
capacitors, diodes and controlled switches. Each of these conventional converters can be
expressed using the basic switching cells and the corresponding circuits are summarized in
the middle column. The converters in this column are made from either N-cell or P-cell. Thus
it is seen that except the boost converter, all of the conventional converters have an inherent
P-cell structure where the active switching element is connected to the positive power supply
terminal. The conventional boost converter is inherently an N-cell boost converter.
Theoretically, all of these classical converters also have a mirror circuit representation.
When the P-cell in a buck converter is replaced with an N-cell, the circuit takes a different
configuration. In this way, the classical boost converter can be re-constructed using a P-cell,
rather than an N-cell. The buck and boost converters can be easily decomposed into a P-cell
and N-cell based circuit, respectively. However, this procedure is not so obvious for the
buck-boost and Ćuk (boost-buck) converters; they inherently take the P-cell structure. The
25
mirror circuit representation of each dc-dc converter is shown in the rightmost column of
Figure 3.3.
V inV o u t
IL V i n
V o u tI L
V inV o u t
I L
V i nV o u t
I L
V inV out
IL
V in
V o u tI L
V inV out
V C
IL1 IL2
V i n
V o u tV C
I L 1
I L 2
B uck converter P-cell buck converter
B oost converter N -cell boost converter
B uck-boost converter P-cell buck-boost converter
C uk converter P-cell C uk converter
V in
V o u t
IL
N -cell buck converter
V in
V o u tI L
P-cell boost converter
V in
V o u t
I L
N -cell buck-boost converter
V in
V o u t
V C
I L 2
IL 1
N -cell C uk converter
(a) Classical dc-dc converters, (b) Formation by the basic cells, (c) Mirror circuits.
Figure 3.3. Four classic DC-DC converters and their P-cell and N-cell representation
To validate the concept of the P-cell and N-cell mirror relationship, a buck converter was
simulated and tested under continuous and discontinuous conduction mode. The simulations
were done in PSIM, and the results are shown in Figure 3.4. Actually, there was no
difference found in the simulation results, implying that there is a mirror relationship
between the N-cell and P-cell structures. Then for further verification, a pair of buck
converters (one P-cell and one N-cell) were constructed from discrete components and tested
in the lab in continuous conduction mode. The operating and loading conditions of the N-cell
buck converter and the P-cell circuit were the same, but some minor differences were
observed in their output voltage. The test results are shown in Figure 3.5.
26
Time (msec)267.7 267.8 267.9 268.0 268.1 268.2
Vdio
de
(V)
I dio
de
(A)
Vsw
itch
(V
)V
L (
V)
I L (
A)
I sw
itch
(A
)
20
0
1.4
0
20
0
0
1.4
-10
10
0
1.5
0
Time (msec)267.7 267.8 267.9 268.0 268.1 268.2
Vdio
de (
V)
I dio
de (
A)
Vsw
itch (
V)
VL (
V)
I L (
A)
I sw
itch (
A)
20
0
1.4
0
20
0
0
1.4
-10
10
0
1.5
0
(a) N-cell buck converter at continuous (b) P-cell buck converter at CCM
conduction mode (CCM)
Vdio
de (
V)
I dio
de (
A)
Vsw
itch (
V)
VL (
V)
I L (
A)
I sw
itch (
A)
20
0
0.7
0
20
0
0
0.7
-10
5
0
0.7
Time (msec)
267.7 267.8 267.9 268.0 268.1 268.2
0
Vdio
de (
V)
I dio
de (
A)
Vsw
itch (
V)
VL (
V)
I L (
A)
I sw
itch (
A)
20
0
0.7
0
20
0
0
0.7
-10
5
0
0.7
Time (msec)
267.7 267.8 267.9 268.0 268.1 268.2
0
(c) N-cell buck converter at discontinuous (d) P-cell buck converter at DCM
conduction mode (DCM)
Figure 3.4. Simulation results for P-cell and N-cell buck converter.
The parameters of the test setup are as follows: Vin = 20 V, D = 0.4, fS = 10 kHz, C1 =
27
100 μF, L1 = 1 mH, D1 = MURB1020CT-1, S1 = IRG4BC30U and RL = 20 .
For an input voltage of 20 V and duty cycle of 0.4, the dc output voltage for the N-cell
structure was 6.82 V and for the P-cell buck converter, it was 7.07 V. Figure 3.5 (a) and (b)
show the output ripple components of the P-cell and N-cell structures respectively. The
fundamental frequency component present in the ripple was the same for both topologies.
However, the N-cell structure produces a cleaner output because of the ground-referenced
gate drive circuit.
(a) P-cell (b) N-cell
Figure 3.5. Experimental output voltage ripple (100mV/div) of buck converter.
The conventional Ćuk converter has an especially unique structure [23]. A Ćuk
converter has continuous input and output current, and the energy is transferred from the
input to the output side by means of a capacitor. The classical Ćuk converter has an inherent
P-cell structure, and the N-cell Ćuk converter can also be achieved. A Ćuk converter is
shown in Figure 3.3 (a), and the switching cell realization is shown in Figure 3.3 (b). The
main limitation of the Ćuk converter is that it uses one additional inductor and capacitor.
However, simplification can be done using the basic switching cells and a new version of
28
Ćuk converter can be obtained. In Figure 3.3 (a), during the time when S1 is on, the rate of
change of currents in L1 and L2 is the following [34]:
dIL1/dt = Vin/L1 (1)
dIL2/dt = [– Vout – (– VC)] /L2 = (VC – Vout) / L2 (2)
where Vout = (ton / toff)Vin (3)
and
VC = (T / toff)Vin (4)
Inserting (4) into (2), we get
dIL2 / dt = (1 / L2)[(TVin – tonVin) / toff] = Vin / L2 (5)
Using the same procedure, the rate of change of currents in L1 and L2 can be found while
S1 is off. Specifically, when S1 is off,
dIL1 / dt = – (1 / L1)(ton / toff)Vin (6)
dIL2 / dt = –(1 / L2)(ton / toff)Vin (7)
Thus, from (1) and (5) - (7), it is concluded that if L1 = L2, the rate of change of currents
in L1 and L2 are the same. Moreover,
IL1(avg) / IL2(avg) = Iin / Iout = D / (1 – D) (8)
From (8), it is found that, for a specific case when the duty ratio D is 0.5, both inductors
will have the same average value of current. If L1 = L2, they will have the same current slope.
Hence, the two inductors can be equivalently moved to the center rail and consolidated into
one inductor. If the converter is not operating at D = 0.5 or if L1 L2, there will be a current
29
mismatch between L1 and L2, and the new Ćuk converter configuration will perform slightly
differently from the original Ćuk converter. Figure 3.6 shows the output voltages of these
converters for a 20 resistive load with a supply voltage of 20 V. The duty cycle of the gate
drive was kept at approximately 0.33, and for this duty cycle, the output voltage of a Ćuk
converter should be around 10 V. In Figure 3.6 (d) - (f), the output ac ripple is shown by
zooming the dc output voltage.
Figure 3.6 shows that these three converters are fairly equivalent. For the same duty
cycle, the P-cell and the N-cell structures produce a 10.6 V dc output, while the new
combined inductor topology produces 10.1 V dc output. These are shown in Figure 3.6 (a), (b)
and (c) respectively. The ripple component in the N-cell circuit has the lowest amplitude of
220 mVp-p, compared to the P-cell structure producing 270 mVp-p. However, the new
topology with the two combined inductors produces a ripple of 340 mVp-p, which is slightly
higher than the other two topologies. Figure 3.6 (d)-(f) show the ripple components in the
three configurations.
30
(a) P-cell Ćuk converter dc output voltage (d) Output ripple of P-cell Ćuk converter
(b) N-cell Ćuk converter dc output voltage (e) Output ripple of N-cell Ćuk converter
(c) New Ćuk converter dc output voltage (f) Output ripple of new Ćuk converter
Figure 3.6. Experimental output voltage of Ćuk converters.
3.1.3 Constructing Voltage Source Inverters from the P-cell and N-cell
Like the dc-dc converters, inverters can be constructed by the use of basic cells in a
similar way. Figure 3.7 (a) shows that the parallel combination of the P- and N- cells creates
31
a phase-leg providing bi-directional current flow. Figure 3.7 (b) shows the conventional anti-
parallel diode/transistor configuration to create a bi-directional current flow. The parallel
connection of a P-cell and an N-cell shown in Figure 3.7 (a) has some distinct advantages
over the conventional IGBT with an anti-parallel diode.
Figure 3.7. Topologies of phase-legs: (a) An inverter phase-leg with bidirectional current flow by
paralleling the P- and N-cells, (b) Conventional connection of anti-parallel diode, (c) Placing two
inductors between P-cell and N-cell common terminals to control the current.
To create a bi-directional current port in a VSI, two transistors in a phase-leg are
switched periodically. However, there is a requirement of dead time between the switching
periods of the two transistors that prevents a short circuit of the dc link. When an inductor is
placed in the paralleled P-cell and N-cell configuration, it takes the shape of Figure 3.7 (c). In
this case, a dead time is not required because the additional inductor and the stray inductance
of the interconnections limit the current if there is any overlap in the switching of the P-cell
and N-cell devices. Therefore, IGBT-diode modules configured as the P- and N-cell are
i
P-cell N -cell
(a)
i
P-cell N -cell
(b)
iL
1L
1
P-cell N -cell
(c)
32
better suited for inverter operation, and at any instant of time, the load current only goes
through the P-cell during the positive half cycle and through the N-cell during the negative
half cycle of the current. Moreover, for a modulation scheme that can detect the direction of
current to the load, only the switch that provides the current path needs to be switched while
the other can be kept off. In the VSI circuit shown in Figure 3.7 (b), when the current is
going to the load, the transistor in the P-cell is switched on and the transistor in the N-cell is
kept off. In the same way, when the current is coming back from the load, it flows through
the transistor in the N-cell which is switched on, and the transistor in the P-cell is kept off.
The basic switching cell concept creates a new vision to analyze the conventional power
electronic converters by segregating them into smaller modular blocks. This modeling
approach is not limited to the use of basic switching cells for analysis of existing power
electronic circuits. Rather, it is a means to find different modular patterns in power electronic
circuits, which can lead to several new circuit topologies.
3.2 Package Layout Design Using the Novel Switching Cells
As discussed above, a P-cell and a N-cell can construct a phase-leg that has some
benefits compared to the traditional anti-parallel phase-leg. Figure 3.8 shows the diagrams of
the two different inverter configurations, one uses conventional phase-leg while the other
uses P-cell and N-cell phase-leg. Under inductive load condition, current commutation is
between S1 and D2 as shown in Figure 3.8(a) when current direction is from load terminal P
to N, or between S2 and D1 when current is from N to P. Therefore, in terms of natural
current commutation pass, it is more reasonable to construct a phase-leg by P-cell and N-cell,
as shown in Figure 3.8(b). Load current flows into the phase-leg through an N-cell and goes
out of the phase-leg through a P-cell. Figure 3.8 also shows the stray inductance within each
33
phase-leg module. This stray inductance model is referred from [37]. L1U and L2L are
introduced by terminal leads; L1L and L2U are the stray inductors of internal bus connecting
the upper and lower unit; the values of these four inductors are relatively large. LC1, Le1, LC2
and Le2, are associated with the die and wire bond, which are relatively small. Since the
physical distance between the two commutating devices is reduced, the inductance is thus
reduced. Comparing Figure 3.8 (a) and (b), inductances L1L and L2U introduced by the
internal bus for the left phase-leg are reduced in the cell structure.
(a) Conventional full bridge inverter (b) Proposed phase-leg
Figure 3.8. Full bridge inverter with package parasitics.
The switching cells are the basic building blocks of almost all power electronics
converters in terms of topology characteristic and operating unit, and they should be used as
the base for manufacturing layout of single, dual and 6-pack modules that semiconductor
manufacturers are produce.
Building power modules and verifying the concept proposed in last section are
extremely expensive and time consuming. However, this process can be simplified by the aid
of simulation. Also, modifying and optimization of the design can be carried out. Ansoft
Q3D Extractor is used to do this job. The software uses Method of Moments (integral
DC
LC1 LC3
LC2 LC4
Le1
Le2
Le3
Le4
Rload Lload
S1
S2
S3
S4
D1
D2
D3
D4
L1U
L2U
L3U
L4U
L1L
L2L
L3L
L4L
P NDC
LC1 LC3
LC2 LC4
Le1
Le2
Le3
Le4
Rload Lload
S1
S2
S3
S4
D1
D2
D3
D4
L1U
L2L
L3U
L4L
P N
P-cell N-cell
34
equations) and Finite Element Method to compute capacitance, conductance, inductance and
resistance matrices. Providing the correct dimensions, material properties (resistivity of
conductors and permittivity of insulators) and boundary conditions (the conductors and
current paths), the software can extract the structural impedances of any arbitrary geometry.
Thus, the module parasitics can be understood thoroughly before modules are fabricated.
3.2.1 Electromagnetic Simulation Methodology
Q3D Extractor is software from Ansoft. It conducts electromagnetic field simulation
employing a combination of the finite element method and the method of moments. In
general, the finite element method divides the full problem space into smaller regions
(tetrahedral elements) and represents the field in each sub-region with a local function. The
method of moments divides the surface (or volumes) of conductors and dielectrics into many
triangular (or tetrahedral) elements to represent the charges and currents present [38].
A finite element solver stores the value of a field quantity (such as electric potential) at
each mesh node of a triangular or tetrahedral element. Inside each element, the field is
interpolated from the values stored at the mesh nodes using local finite element basis
functions. Maxwell‟s partial differential equations of the field quantities are linearized and
can be transformed into a sparse matrix of linear algebraic equations that can be solved using
traditional direct or iterative numerical methods. In the method of moments, field quantities
are also interpolated over elements, but typically with simpler basis functions than in the
finite element method. For capacitance problems, the field quantity of interest is the charge
density on the surface of a conductor or dielectric interface. Triangular elements are used,
and the charge density is approximated with a piecewise-constant basis function. For
inductance problems, the field quantity is a vector (current density), and again piecewise-
35
constant basis functions are used. For DC inductance problems, the elements are tetrahedra
(volume currents), and for AC inductance problems the elements are triangles (surface
currents.) Green‟s function is used to represent the electrical interaction between any pair of
elements. Using integrals of the Green‟s function and the basic functions, a dense matrix of
linear equations is derived. An iterative method is used solve large problems.
The procedures of conducting the simulation are described in the following subsection.
3.2.2 Layout Design Using Q3D Extractor
A. Build the three dimensional model of the power device module and specify material
The three dimensional model can be drawn in Q3D Extractor or imported from other
software, such as Auto CAD or Protel. Any of the following formats can be opened and
edited: 3D Modeler file (*.sm3), SAT file (*.sat), STEP file (*.step,*. stp), IGES file (*.iges,
*.igs) and ProE files (*.prt, *.asm) in Q3D Extractor.
The proposed phase-leg and the conventional phase-leg are drawn using Q3D Extractor.
The diagrams of the phase-legs are shown in Figure 3.9, the marked loop1 and loop 2 are two
current commutation loops in a phase-leg, that is, from upper IGBT to lower diode and from
lower IGBT to upper diode. In a conventional module, the upper leg devices S1 and D1 are
seated at one side, while the lower leg, namely S2 and D2 are seated at the other side. The
layout designs are shown in Figure 3.9. The purpose here is to compare the stray inductance
in the modules, therefore the base plate, case and encapsulant, which do not affect the
electrical characteristic of the module, are neglected in the drawings. In the three dimensional
modules, the bottom layer is DBC, the semiconductor dice are seated on the DBC, the black
lines represent the bonding wires. The terminals are marked as in the figures. The physical
36
distance for loop1 is shown as the red trace in Figure 3.9 (b). It starts from lead C1 and
passes through S1, two groups of bond wires, output bus E1C2 and more wires to D2. Loop2
is shown as the green line, and the length for loop2 is similar with loop1. In the proposed P-
cell and N-cell modules in Figure 3.9 (b), the two devices in the commutation loop are seated
at the same side. Thus, the physical length of the commutation loop is reduced. For example,
loop1 shown as the red trace also starts from C1, goes through only one group of wires, and
then reaches D2. This is much shorter than the same loop in a conventional module. For a
better comparison purpose, the two modules are similar in terms of substrate size and lead
frame position. The physical size is listed in Table 3.1.
After the geometries of the modules are built in Q3D Extractor, materials for each of the
components are assigned, which are shown in Table 3.2.
B. Analysis Setup
The next step is to assign excitations, which includes source and sink, representing the
current in terminal and current out terminal, respectively. The parasitics of the path between
source and sink is calculated. If there are multiple paths, the mutual parasitics between the
paths will also be calculated. The results are stored in a matrix. Specifically, the items in the
diagonal line are the self inductance (capacitance or resistance), and others are mutual values.
It is necessary to point out that the size of the excitations can affect the results. For example,
the larger the area of the excitations, the smaller the parasitic inductance is. Although it is not
a dominant factor, to have the similar condition, it is better to draw the area as the real
excitation contact area in the measurement.
When setting up the solver, there are options for DC or AC excitation source. When
37
using DC excitation, current is distributed evenly in the geometry, while using AC excitation,
skin effect is considered. The default frequency for AC excitation is 100 kHz, the parasitic
inductance is not affected when changing frequency, and capacitance does not change either;
only resistance is scaled with the factor of (fnew/fold)1/2
. For inductance, which is the focus in
this work, the DC result is larger than the AC result because the volume considered in DC
simulation is larger.
(a) Conventional phase-leg module
(b) Proposed phase-leg module
Figure 3.9. Phase-leg module layout.
38
Table 3.1. Power Module Components Dimensions
Conventional module Proposed module
DBC size (mm) 37.0×38.0 37.5×38.5
DBC thickness (mil) 8(Cu), 25(Alumina) 8(Cu), 25(Alumina)
IGBT (mm) 5×5 5×5
Diode (mm) 5.85×5.85 5.85×5.85
Bond wires (diameter×number) 8 mil×5 8 mil×5
Table 3.2. Power Module Materials
Components Materials
DBC insulator Alumina, permittivity 9.2
DBC metallization Copper, conductivity 58000000S/m
Dice Silicon, permittivity 11.9
Bond wires Aluminum, conductivity 38000000S/m
3.2.3 Simulation Results
The parasitic inductance associated with the module is studied thoroughly. The source
and sink points are shown in Figure 3.10 as the yellow area. The simulation results are listed
in Table 3.3.
The path from point A to point B is the commutation loop from the positive bus to the
negative bus. There are two conduction paths: through S1-D2 which is loop1 shown in
Figure 3.9, and through S2-D1 which is loop2. The simulation is conducted this way: when
calculating loop1, the materials of S1 and D2 are set to copper while S2 and D1 is set to
silicon, so that only the S1-D2 path conducts; the other path is calculated the same way. The
inductance of the bus bar is also calculated separately. The path from A to D is the positive
39
bus bar, the path from B to E is the negative one, and the path from C to F is the AC output
bus bar. They should have the same value since the same structure is used. In a commercial
module, bus bars usually have larger dimensions and complex shape, however to simplify the
fabrication process, the simulation uses a simpler copper bar. From H to G and from J to I are
the gate drive loops.
It can be seen that the commutation loop inductances in the conventional module are
both around 20 nH, while in the proposed module they are less than 10 nH. As expected, the
rearrangement of the dice layout has an obvious effect on reducing the inductance of the
DBC trace. For the other values, there are no large differences.
(a) Conventional module (b) Proposed module
Figure 3.10. Phase-leg modules with measuring points.
40
Table 3.3. Simulation Results of the Phase-leg Module Stray Inductance
Conventional Layout Proposed Layout
DC Value (nH) AC Value (nH) DC Value (nH) AC Value (nH)
A-B (Loop1) 38.3 18.0 11.7 7.2
A-B(Loop2) 38.3 18.5 14.4 8.6
A-D (also B-E or
C-F)
4.2 3.5 4.2 3.5
G-H 16.6 13.5 16.6 13.5
I-J 16.6 13.5 16.6 13.5
E-C (via S2) 24.4 16.7 18.4 12.4
E-C (via D2) 21.4 14.6 15.8 11.3
C-D (via D1) 24.3 15.9 24.4 15.8
C-D (via S1) 27.1 20.8 20.9 14.7
3.3 Summary
This chapter first introduces the concepts of P-cell and N-cell, and how they work as the
basic function units in power converters. From packaging view point, the novel switching
cells make it convenient to build power converters, by reducing the commutation loop
parasitic inductance compared to the anti-parallel cells. Based on the topology analysis, a
new packaging layout design of phase-leg module based on the novel switching cells is
proposed. The proposed layout design as well as a conventional design is built using Ansoft
Q3D Extractor. Electromagnetic simulation is conducted to study the module package
parasitics. The detailed stray inductances inside the phase-leg modules are extracted. The
simulation results show that the commutation loop inductance in the proposed module is
largely reduced compared to the conventional module. This can benefit the IGBT switching
behavior, which will be discussed in the next chapter.
41
4 Electrical Evaluation Based on Simulation
After the extraction of the module parasitics, mainly the stray inductance, electrical
evaluation is performed under the influence of these parasitic elements for the two different
power module layout cases. In power electronics, the most common actions are turning off
and turning on of a switch. Thus the switching characteristic has been chosen to identify the
performance of the two modules. In order to test the switching characteristic, the detailed
parasitics in a test circuit has to be identified, which are mainly the stray inductance of the
circuit trace and the parasitic capacitance of the power semiconductor devices. Fabricating a
power module and building testing circuits are time consuming, but simulation can greatly
accelerate the process. This chapter discusses the switching characterization using Synopsys
Saber.
4.1 Parasitics Extraction from PCB
As mentioned previously, the DC bus connection also brings parasitic inductance to the
commutation loop that cannot be neglected. The DC power comes from the power supply
and a large aluminum electrolytic capacitor, and then goes to the PCB. Usually, several low
ESR decoupling film capacitors are used to compensate the ESL of the cable. Therefore only
the inductance from the capacitor to the DC bus needs to be considered. The printed circuit
board (PCB) is shown in Figure 4.1. The decoupling capacitors and the power module are
shown in the figure. The traces (polygon) between them are the ones which should be
counted for the parasitics. To estimate the value, this part is analyzed using Q3D Extractor.
Specifically, the top and bo
Recommended